103 A publica ation of CHE EMICAL ENGINEER RING TRAN NSACTION NS VOL.. 37, 2014 Guest E Editors: Eliseo Ran nzi, Katharina Kohse- Höinghaus Copyrig ght © 2014, AIDIC C Servizi S.r.l., ISBN 9 978-88-95608-28-0; ISSN 2283-9216 The Italian Asso ociation oof Chemical Engin neering www.aidic.it/cet DOI: 10.3303/CET1437018 Statistical Description of Biomas ss Blend ds Devollatilizatio on Jaku ub Bibrzyckki, Anna Ka atelbach-W Woźniak*, Magdalena M a Niestrój, Andrzej Szlęk Silesia an University of Technology, Ins stitute of Therm mal Technology, Gliwice, Poland d [email protected] Bioma ass is widely used as a renewable enerrgy source eitther as a raw material or aas a form of pellets, p which are very oftten produced of mixture o of different sorts of biomas ss. This givess the opportunity to cterized by de esired propertiies. However there is a queestion of intera actions produce pellets which are charac een sorts of biomass which may occur du uring pyrolysis s. The main sc cope of this paaper is to investigate betwe devola atilization of biomass mixtture and to a approximate amount a of released volatilees as a function of bioma ass compositio on. Five types of bioma ass has been selected for m measurements s: two types of wood – oak and pine, two o types eat straw and d willow as popular energyy crop. It has s been of agrricultural wastes – rape sttraw and whe assum med that the amount of vo olatiles releassed in a given temperature e range can be expressed d as a functio on of elementtal compositio on of biomasss mixture. Diffferent forms of o function hav ave been teste ed and correlation coefficie ents as well as average an d maximal ap pproximation errors e were exxamined as defining uality of apprroximation. It has been fou und that linea ar form of a function givess the best qua ality of the qu appro oximation charracterized with h high correla ation coefficient and relatively small maxximal approximation error. 1. Inttroduction Limite ed resources of fossil fuels s as well as problem of global g climate changes aree the motivation for increa asing role of renewable r ene ergy sources. For many co ountries bioma ass is the onlyy renewable energy e source e which may play a signific cant role. For e example Pola and has relativ vely low wind vvelocities at most m of the te erritory, flat area a and ave erage solar ra adiation and thus wind turbines, hydroo power plantts and photovoltaic source es cannot play y a significantt role and the best option for increasing share of rene ewable gy sources is to o energy utiliz zation of bioma ass. energ ass can be ussed either as a raw material such as straw w, wood-chips s etc. or in a pprocessed form m such Bioma as forr example pellets and brique ettes. In the s econd case during production of pellets aand briquettes s there is opportunity for selecting bio omass mixture e in a prope er way in ord der to obtainn pellets whic ch are d properties. When u using mixture es of differentt kinds of bioomasses therre is a characterized by desired ay occur durin ng the thermal conversion off biomass. question of interactions which ma he other hand, some resea archers (Wang g, et al., 2011 1; Couhert, ett al., 2009) haave an opinio on that On th pyrolyysis of biomasss cannot be predicted p base ed on composition due to ex xisting interacttions. Giudicia anni, et al. (20 013) carried out o steam pyro olysis of two-ccomponent mix xtures and compared outcoomes to resultts from calculations using additive a law. Authors A notice ed differences in product dis stribution, gas composition as a well as HH HV of gas. The largest deviiation was dettected in case e of xylan-lignin mixture, whhat is in accord with resultss from (Liu, ett al., 2011). In n another worrk (Hosoya, ett al., 2007), py yrolysis of celllulose-hemice ellulose and ccellulose-lignin n mixtures we ere investigatted. Strong in nteractions be etween celluloose and lignin n were obserrved, which led to product composition c d differences. Th hese observations are on tthe contrary to o work (Wang g, et al., 2011 1), where only y weak interacction in the mixture m was no oticed. Authorss of work (Wa ang, et al., 20 011) indicated d that interactiions in mixturre of cellulose e-hemicellulose e were marginnal, what is in n good accord dance with oth her researche es (Giudiciannii, et al., 2013; Hosoya, et al., 2007). Inconssistency in intteraction occu urrence betwe en major biom mass organic components ccan be explain ned by hemiccellulose and lignin differe ent chemical fforms presen nce in biomas ss, therefore products yield and interactions betwee en them can be e different. Please cite this article as: Bibrzycki J., Katelbach-Wozniak A., Niestroj M., Szlek A., 2014, Statistical description of biomass blends devolatilization, Chemical Engineering Transactions, 37, 103-108 DOI: 10.3303/CET1437018 104 The main objective of this paper is to search for the correlation between amount of volatiles released by the biomass mixture and properties of single biomasses which were used for mixture preparation. 2. Experimental investigation Five different types of biomass have been selected for measurements, among them two wood biomass: oak and pine, two agricultural by-products: rape straw and wheat straw and one energy crop - willow. Elemental composition as well as proximate analysis of these biomasses is presented in Table 1 and ash composition in Table 2. Table 1: Properties of biomasses used in experiments parameter unit rape straw wheat straw willow pine oak water %,mass 2.3 5.1 2.8 3.0 1.3 ash %,mass 4.0 5.7 2.3 0.4 0.2 volatiles %,mass 76.8 71.5 77.7 82.0 80.7 LCV MJ/kg 17.065 16.331 17.519 18.037 18.079 carbon %, mass 46.64 43.92 47.69 49.11 49.13 hydrogen %, mass 5.98 5.49 5.90 6.13 5.90 nitrogen %, mass 0.66 0.99 0.34 0.01 0.04 sulfur %, mass 0.16 0.14 0.04 0.02 0.03 chlorine %, mass 0.046 0.129 0.004 0.003 0.005 fluorine %, mass 0.004 0.004 0.000 0.001 0.004 Table 2: Biomass ash composition Si, % Ca, % Mg, % S, % P, % K, % 29.29 9.32 3.95 4.80 7.55 33.0 rape straw 5.31 34.30 3.06 6.74 6.48 17.6 willow 2.14 39.40 3.45 2.53 6.12 14.10 pine 24.50 23.50 5.90 3.67 3.15 11.60 oak 6.64 17.30 3.12 3.82 3.97 34.40 wheat straw For each of the pure components as well as for the mixture of biomasses in a proportion 30 %/ 70%, 50 %/50 % and 50 %/70 % TGA tests were done using Netsch thermo-balance, nitrogen as neutral atmosphere and heating rate 10 K/min as typical for fixed bed combustion. An example of a result of such measurement is shown in Figure 1. 105 tx 100 0 DTG TG, % 80 -4 60 40 -8 20 t, C 0 -12 100 200 300 400 500 600 Figure 1: Example TG and DTG as a function of temperature for sample – oak, atmosphere-nitrogen, heating rate 10 K/min In the Figure 1 it is shown characteristic temperature t=300 oC which is a temperature for which most of the hemicellulose is already decomposed while cellulose and lignin still not. In total 49 different mixtures were investigated. Approximation of results It was assumed that the total amount of volatiles organic part of mixture: v should be a function of elemental composition of an v = f ( c , h, o ) where c , h, o (1) denote mass fraction of respectively carbon, hydrogen and oxygen in a organic matter. Since in organic matter v = f (c, h ) c + h + o ≈ 1 it can be written that: (2) It has been also assumed that the function should be polynomial of the order no greater than 2. Under such conditions eight functions can be written which are presented in Table 3. For each of the functions least square procedure was applied obtain the coefficients and next it was calculated average error of approximation, maximal error of approximation and correlation coefficient of measurement values with values calculated using tested functions. Results are shown in Table 4. 106 Table 3: Formulas for approximation functions which were tested Formula Number of function Number of coefficients 1 3 a1 + a2 c + a3 h 2 4 a1 + a2 c + a3 h + a 4 c 2 3 4 a1 + a2 c + a3 h + a 4 h 2 4 4 a1 + a2 c + a3 h + a 4 ch 5 5 a1 + a2 c + a3 h + a 4 h 2 + a5 c 2 6 5 a1 + a2 c + a3 h + a4 h 2 + a5 ch 7 5 a1 + a2 c + a3 h + a 4 c 2 + a5 ch 8 6 a1 + a2 c + a3 h + a 4 c 2 + a5 h 2 + a6 ch Table 4: Average error, maximal error and correlation coefficient of tested functions 1. Function Avg error, % Max error, % Correlation 1 0.83 2.02 0.947 2 0.80 2.30 0.958 3 0.82 2.28 0.957 4 0.83 2.23 0.958 5 0.77 2.30 0.958 6 0.75 2.22 0.957 7 0.76 2.28 0.958 8 0.67 2.27 0.958 It can be noticed that with increasing number of coefficients in a function average error decreases which is a rule for approximation problems. However, at the same time maximal error increases with increasing number of coefficients, while correlation coefficient remains almost constant. Dependence of the maximal and average errors and correlation coefficient on the number of coefficients in a function is shown in Figure 2. Basing on this consideration it can be concluded that the best function of all tested is the one which has a form: v = a1 + a2 c + a3 h (3) or, for a given set of biomasses: v = −14,2917 + 116,44c + 538,74h (4) 107 Similar considerations were repeated for the amount of volatiles released in a temperature range up to 300 oC. Also in this case the best function was the one having only linear expression. For the amount of o volatiles released in the temperature range up to 300 C function has following form: v300 = 153,8 − 0,7585c − 16,122 h 2.5 (5) δsrδmax 1.00 δmax R 2.0 0.98 R 1.5 1.0 0.96 0.94 δsr 0.5 0.92 0.0 3 4 5 n 0.90 6 Figure 2: Maximal and average error as well as correlation coefficient R as a function of number of functions coefficients n 3. Summary In the paper, it is shown how different forms of approximation function can describe amount of volatiles releases from the biomass mixture. It has been assumed that amount of volatiles can be expressed as a function of elemental composition of organic matter of a biomass mixture. Eight different forms of function were tested and finally the simplest one was concluded to be the best one having relatively small maximal error of approximation and good enough correlation coefficient. Consideration was done separately for total amount of volatiles as well for the amount of volatiles released o in the temperature range up to 300 C. The form of the best fitting function was the same in both cases while the coefficients in both functions were different. Obtained results show the possibility of using approximation functions for proper selection of biomass types which should be used to obtain the mixture of desired properties. This can be used for example for pellets production. 4. Acknowledgments This work has been supported by the Polish National Science Centre, project number N N513 325740 “Pyrolysis of biomass mixtures”. References Bernhard P., 2011, Prediction of pyrolysis of pistachio shells based on its components hemicelluloses, cellulose and lignin, Fuel Processing Technology, Vol. 92, 1993-1998. Couhert C., Commandre J., Salvado S., 2009, Is it possible to predict gas yields of any biomass after rapid pyrolysis at high temperature from its composition in cellulose, hemicellulose and lignin?, Fuel, Vol. 88, 408-417. 108 Eom I., Kim J., Kim T., Lee S., Choi D., Choid I., Choi, J, 2012, Effect of essential inorganic metals on primary thermal degradation of lignocellulosic biomass, Bioresource Technology, Vol. 104, 687-694. Giudicianni P., Cardone G., Ragucci R., 2013, Cellulose, hemicellulose and lignin slow steam pyrolysis: Thermal decomposition of biomass components mixtures, Journal of Analytical and Applied Pyrolysis, Vol. 100, 213-222. Han L., Wang Q., Ma Q., You C., Lupo Z., Cen K., 2010, Influence of CaO additives on wheat-straw pyrolysis as determined by TG-FTIR analysis, Journal of Analytical and Applied Pyrolysis, Vol. 88, 199-206. Hosoya T., Kawamoto H., Saka S., 2007, Cellulose-hemicellulose and cellulose-lignin interaction in wood pyrolysis at gasification temperature, Journal of Analytical and Applied Pyrolysis, Vol. 80, 118-125. Liu Q., Zhong Z., Wang S., 2011, Interactions of biomass components during pyrolysis: A TG-FTIR study, Journal of Analytical and Applied Pyrolysis, Vol. 90, 213-218. Wang S., Guo X., Wang K., Luo Z., 2011, Influence of the interaction of components on the pyrolysis behavior of biomass, Journal of Analytical and Applied Pyrolysis, Vol. 91, 183-189. Yang H., Yan R., Chen H., Zheng C., Lee D., Liang D., 2006, Influence of mineral matter on pyrolysis of palm oil wastes, Combustion and Flame, Vol. 146, 605-611.